Estimation method, printing method, and printing device

Abstract

To provide a technique capable of estimating an expansion / contraction state of a base material according to submitted data before printing.SOLUTION: A printer includes a conveyance mechanism 10, heads 21 to 24, and an estimation part 44. The conveyance mechanism 10 conveys a long belt-like base material in a longer direction along a predetermined conveyance path. The heads 21 to 24 discharge ink onto a surface of the base material conveyed by the conveyance mechanism 10 on the basis of submitted data D1. The estimation part 44 outputs an estimation result D5 showing an expansion / contraction state of the base material by the ink on the basis of the submitted data D1 before printing of the submitted data D1. Thereby, the estimation result D5 of the expansion / contraction state according to the submitted data D1 can be obtained. Therefore, the submitted data D1 can be printed in consideration of the estimation result D5.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to a technique for estimating the expansion and contraction state of a long strip-shaped substrate in a printing device that ejects ink onto the surface of the substrate while transporting the substrate in the longitudinal direction. [Background technology]

[0002] Conventionally, there is known an inkjet printing device that prints an image on a long strip-shaped substrate by ejecting ink from a plurality of heads while conveying the substrate in the longitudinal direction. The inkjet printing device ejects ink of different colors from a plurality of heads. Then, a multi-color image is printed on the surface of the substrate by superimposing monochromatic images formed by the inks of the respective colors. A conventional printing device is described, for example, in Patent Document 1. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2004-268361 A Summary of the Invention [Problem to be solved by the invention]

[0004] In this type of printing device, when the substrate absorbs ink or when the ink on the substrate dries, the substrate expands and contracts slightly. As a result, the printed image formed on the surface of the substrate 9 is distorted. In addition, the expansion and contraction of the substrate may cause the ink ejected from the multiple heads to shift position. Conventionally, various parameters of the printing device have been set to minimize the deterioration of print quality caused by the expansion and contraction of the substrate.

[0005] However, the amount of expansion and contraction of the substrate varies depending on various conditions such as the tension applied to the substrate, the type of substrate, and the type of ink. In addition, even between printing devices of the same type, the amount of expansion and contraction of the substrate may differ due to differences in the devices. For this reason, it has been difficult to set the above parameters uniquely. In addition, the expansion and contraction state of the substrate differs depending on the image data to be printed, but conventionally, it has not been possible to know the expansion and contraction state of the substrate before printing.

[0006] The present invention has been made in consideration of the above circumstances, and has an object to provide a technique that can estimate the expansion / contraction state of a base material according to input data before printing. [Means for solving the problem]

[0007] In order to solve the above problems, a first aspect of the present invention is a method for estimating an expansion / contraction state of a long strip-shaped substrate in a printing device that ejects ink onto a surface of the substrate while transporting the substrate in a longitudinal direction, the method comprising: a learning process for generating an estimation model capable of estimating the expansion / contraction state of the substrate by machine learning using learning data, which is image data for learning, as input variables and the expansion / contraction state of the substrate when the learning data is printed by the printing device as training data; a data acquisition step of acquiring submitted data, which is image data to be printed; and an estimation step of outputting an estimation result indicating an expansion and contraction state of a base material due to ink, based on the submitted data, prior to printing of the submitted data, The learning data is image data incorporating a plurality of marks arranged at intervals, and in the learning data, the marks are arranged more densely in portions where there is a large change in density value or a large change in printing rate than in other portions. In the learning process, an expansion / contraction state serving as the teacher data is measured based on positions of the plurality of marks on the substrate when the learning data is printed by the printing device. In the estimation process, the input data is input to the estimation model, and the expansion / contraction state output from the estimation model is regarded as the estimation result. The estimation result is an estimation of the two-dimensional distribution of the amount of expansion and contraction in the transport direction and width direction of the base material.

[0010] The present application 2 The invention is 1 An estimation method of the invention, wherein the mark includes a base figure formed in a first color and a mark figure formed in a second color different from the first color at a position overlapping the base figure.

[0011] The present application 3 The invention is 2 In the estimation method of the invention, the first color is white and the second color is black.

[0012] The present application 4 The invention is 1 From the invention 3In the estimation method of any one of the above aspects, the teaching data indicates a change in the distance between adjacent marks.

[0013] The present application 5 The invention is 1 From the invention 3 In the estimation method of any one of the above aspects, the teaching data indicates a displacement amount of each of the plurality of marks.

[0014] The present application 6 The invention is 1 From the invention 5 In the estimation method of any one of the above aspects of the invention, the learning data is image data including a plurality of areas with different density values ​​or printing rates.

[0015] The present application 7 The invention is the first invention to the second invention. 6 A printing method using the estimation method of any one of the inventions up to the present invention, wherein after executing the data acquisition process and the estimation process, a printing process is executed in which ink is ejected onto a surface of the substrate while correcting the ink ejection position relative to the substrate based on the estimation result.

[0016] The present application 8 The invention is a printing device, comprising: a data acquisition unit that acquires input data, which is image data to be printed; a transport mechanism that transports a long strip-shaped substrate in a longitudinal direction along a predetermined transport path; and a head that ejects ink onto a surface of the substrate transported by the transport mechanism based on the input data. a camera that photographs a printed surface of the substrate downstream of the head on the transport path; an expansion / contraction state measurement unit that measures an expansion / contraction state of the substrate based on an image photographed by the camera; and a learning unit that uses learning data, which is image data for learning, as input variables and the expansion / contraction state when the learning data is printed as teacher data to generate an estimation model capable of estimating the expansion / contraction state of the substrate by machine learning. an estimation unit that outputs an estimation result indicating a state of expansion and contraction of the base material due to the ink based on the input data before printing of the input data, the learning data is image data incorporating a plurality of marks arranged at intervals, the marks are arranged more densely in portions of the learning data where there is a large change in density value or a large change in printing rate than in other portions, the expansion / contraction state measuring unit measures an expansion / contraction state to be the teacher data based on positions of the plurality of marks on the substrate when the learning data is printed, the estimation unit inputs the submitted data to the estimation model, and the expansion / contraction state output from the estimation model is regarded as the estimation result, The estimation result is an estimation of the two-dimensional distribution of the amount of expansion and contraction in the transport direction and width direction of the base material.

[0018] The present application 9 The invention is 8The printing device of the invention further includes a correction value calculation unit that calculates a correction value based on the estimation result, and an operation control unit that controls the transport mechanism and the head so as to eject the ink onto the surface of the substrate while correcting the ink ejection position relative to the substrate based on the correction value. Effect of the Invention

[0019] The first to third aspects of the present application 9 According to the present invention, the expansion / contraction state of the base material due to the ink is estimated before printing based on the input data. This makes it possible to obtain an estimated result of the expansion / contraction state according to the input data. Therefore, the input data can be printed taking into account the estimated result of the expansion / contraction state. In addition, by using machine learning, it is easier to respond to changes in conditions compared to when the estimated result of the stretching state is calculated based on a formula or table that shows the relationship between image data and the stretching state. Furthermore, by using multiple marks incorporated in the learning data, the stretching state of the base material that serves as training data can be easily obtained.

[0022] In particular, 2 According to the invention, the mark graphic can be identified regardless of the color of the background image.

[0023] In particular, 4 According to the present invention, the training data represents the amount of local expansion and contraction for each region. Since it does not represent the amount of accumulated displacement, it is easy to handle in machine learning.

[0024] In particular, 5 According to the invention, the teacher data represents changes in the coordinate positions of each part of the substrate. In this case, the estimation results output from the estimation model also represent changes in the coordinate positions of each part of the substrate. Therefore, the coordinate positions of the ink ejected onto the substrate can be easily corrected based on the estimation results.

[0025] In particular, 6 According to the invention, it is possible to learn the amount of expansion and contraction of the substrate depending on the density value or the printing rate. [Brief description of the drawings]

[0026] [Figure 1] FIG. 1 illustrates a configuration of a printing device. [Diagram 2]FIG. 2 is a partial top view of the printing device in the vicinity of the printing unit. [Diagram 3] FIG. 2 is a block diagram showing connections between each unit of the printing device and a computer. [Figure 4] FIG. 1 is a block diagram conceptually illustrating the functions of a computer. [Diagram 5] 13 is a flowchart showing the flow of a learning process. [Figure 6] FIG. 11 is a diagram showing an example of learning data. [Figure 7] FIG. 2 is an enlarged view of a portion of the learning data. [Figure 8] 10A to 10C are diagrams illustrating an example of a method for measuring the expansion and contraction state of a substrate. [Figure 9] 10A to 10C are diagrams showing another example of a method for measuring the expansion and contraction state of a base material. [Figure 10] 13 is a flowchart showing the flow of a printing process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0028] <1. Printer configuration> FIG. 1 is a diagram showing the configuration of a printing device 1 according to an embodiment of the present invention. This printing device 1 is an apparatus that prints an image on the surface of a long strip-shaped substrate 9 by ejecting ink droplets from a plurality of heads 21-24 toward the substrate 9 while transporting the substrate 9. The substrate 9 may be printing paper or a resin film. The substrate 9 may also be a substrate made of metal foil or glass. As shown in FIG. 1, the printing device 1 includes a transport mechanism 10, a printing unit 20, a camera 30, and a computer 40.

[0029] The transport mechanism 10 transports the substrate 9 in a transport direction along its longitudinal direction. The transport mechanism 10 of this embodiment has an unwinding section 11, a plurality of transport rollers 12, and a winding section 13. The substrate 9 is unwound from the unwinding section 11 and transported along a transport path formed by the plurality of transport rollers 12. Each transport roller 12 rotates about an axis extending in a direction perpendicular to the transport direction, thereby guiding the substrate 9 downstream of the transport path. After transport, the substrate 9 is collected in the winding section 13. In addition, tension is applied to the substrate 9 in the transport direction. This suppresses sagging and wrinkling of the substrate 9 during transport.

[0030] The printing unit 20 is a processing unit that ejects ink droplets (hereinafter referred to as "ink droplets") onto the substrate 9 transported by the transport mechanism 10. The printing unit 20 of this embodiment has a first head 21, a second head 22, a third head 23, and a fourth head 24. The first head 21, the second head 22, the third head 23, and the fourth head 24 are arranged at intervals along the transport direction of the substrate 9. The substrate 9 is transported below the four heads 21 to 24 with the printing surface facing upward.

[0031] Fig. 2 is a partial top view of the printing device 1 in the vicinity of the printing unit 20. As indicated by dashed lines in Fig. 2, a plurality of nozzles 201 arranged parallel to the width direction of the substrate 9 is provided on the bottom surface of each of the heads 21-24. Each of the heads 21-24 ejects ink droplets of each color, K (black), C (cyan), M (magenta), and Y (yellow), which are color components of a multi-color image, from the plurality of nozzles 201 toward the top surface of the substrate 9.

[0032] That is, the first head 21 ejects K ink droplets onto the upper surface of the substrate 9 at a first printing position P1 on the transport path. The second head 22 ejects C ink droplets onto the upper surface of the substrate 9 at a second printing position P2 downstream of the first printing position P1. The third head 23 ejects M ink droplets onto the upper surface of the substrate 9 at a third printing position P3 downstream of the second printing position P2. The fourth head 24 ejects Y ink droplets onto the upper surface of the substrate 9 at a fourth printing position P4 downstream of the third printing position P3.

[0033] A fixing unit for fixing the ink on the printing surface of the substrate 9 may be further provided downstream in the transport direction of the heads 21 to 24. The fixing unit dries the ink adhering to the substrate 9, for example, by blowing heated gas toward the substrate 9. However, the fixing unit may also cure the ink by irradiating UV-curable ink with ultraviolet rays.

[0034] The camera 30 is an imaging device that captures an image of the printed surface of the substrate 9 that has passed through the printing unit 20. The camera 30 is disposed facing the printed surface of the substrate 9 at an imaging position P5 that is downstream of the four heads 21-24 on the transport path. The camera 30 is, for example, a line sensor in which a plurality of imaging elements such as CCD or CMOS are arranged in the width direction. The camera 30 captures an image of the printed surface of the substrate 9. The camera 30 then transmits the captured image to the computer 40.

[0035] The computer 40 is an information processing device for controlling the printing device 1. Fig. 3 is a block diagram showing the connection between the computer 40 and each unit of the printing device 1. As conceptually shown in Fig. 3, the computer 40 has a processor 401 such as a CPU, a memory 402 such as a RAM, and a storage unit 403 such as a hard disk drive. A computer program 404 for executing a learning process and a printing process, which will be described later, is stored in the storage unit 403.

[0036] As shown in FIG. 3, the computer 40 is communicatively connected to the above-mentioned transport mechanism 10, the four heads 21-24, and the camera 30. The computer 40 is also communicatively connected to a server 2, which is an external storage device. The server 2 stores submitted data D1. The submitted data D1 is image data to be printed by the printer 1 to obtain a printed matter as a product. The computer 40 acquires the submitted data D1 from the server 2, and controls the operation of the transport mechanism 10 and the four heads 21-24 based on the submitted data D1. This allows the printing process in the printer 1 to proceed. The server 2 also stores a large amount of learning data D2 used in the learning process described below.

[0037] <2. Computer functions> In this printing device 1, the four heads 21-24 eject ink droplets to print single-color images on the upper surface of the substrate 9. The four single-color images are then superimposed to form a multi-color image on the upper surface of the substrate 9. Therefore, if the ink droplets ejected from the four heads 21-24 are misaligned from one another on the substrate 9, the image quality of the printed matter will deteriorate. Keeping such misalignment (misregistration) of the single-color images on the substrate 9 within an acceptable range is an important factor in obtaining a high-quality printed matter.

[0038] Furthermore, when the ink is absorbed into the substrate 9 or when the ink on the substrate 9 dries, the substrate 9 expands and contracts non-uniformly. When the substrate 9 expands and contracts in this way, the printed image formed on the printing surface is distorted. When the substrate 9 expands and contracts, the misregistration described above is also likely to occur. Therefore, the computer 40 of this printing device 1 has a function of estimating the expansion and contraction state of the substrate 9 before printing based on the input data D1, and performing printing while correcting the ejection position of ink droplets using the estimation result. FIG. 4 is a block diagram conceptually showing the function of the computer 40.

[0039] 4, computer 40 has a data acquisition unit 41, an expansion / contraction state measurement unit 42, a learning unit 43, an estimation unit 44, a correction value calculation unit 45, and an operation control unit 46. The functions of data acquisition unit 41, expansion / contraction state measurement unit 42, learning unit 43, estimation unit 44, correction value calculation unit 45, and operation control unit 46 are realized by processor 401 of computer 40 operating in accordance with a computer program 404.

[0040] The data acquisition unit 41 is a processing unit for acquiring the submitted data D1 and the learning data D2. The data acquisition unit 41 reads out the submitted data D1 and the learning data D2 from the server 2. The data acquisition unit 41 also inputs the read out learning data D2 to the learning unit 43 and the operation control unit 46. The data acquisition unit 41 also inputs the read out submitted data D1 to the estimation unit 44 and the operation control unit 46.

[0041] The expansion / contraction state measuring unit 42 is a processing unit for measuring the expansion / contraction state of the substrate 9 based on the captured image D3 transmitted from the camera 30. A method for measuring the expansion / contraction state of the substrate 9 will be described later. The expansion / contraction state measuring unit 42 inputs the measured expansion / contraction state D4 of the substrate 9 to the learning unit 43.

[0042] The learning unit 43 is a processing unit for learning the relationship between image data and the expansion / contraction state of the substrate 9 when the image data is printed. In the printing device 1, the learning process is performed by the learning unit 43 before printing of the input data D1 for obtaining a printed matter that will become a product. During the learning process, learning data D2 is input to the above-mentioned operation control unit 46 and the learning unit 43. The operation control unit 46 controls the operation of the transport mechanism 10 and the printing unit 20 based on the learning data D2. As a result, the learning data D2 is printed on the printing surface of the substrate 9. Also, during the learning process, the camera 30 captures an image of the substrate 9 after printing.

[0043] The learning unit 43 performs a learning process by a supervised machine learning algorithm using the learning data D2 as an input variable and the stretch state D4 measured by the stretch state measuring unit 42 as teacher data. The learning unit 43 repeats such a learning process until a predetermined end condition is satisfied. As a result, the learning unit 43 generates an estimation model M that can output an estimation result indicating the stretch state of the base material 9 based on the input image data. The learning process will be described in detail later.

[0044] The learning unit 43 uses, for example, deep learning (multi-layer neural network) as a machine learning algorithm. However, the machine learning algorithm used by the learning unit 43 is not limited to deep learning. Instead of deep learning, other machine learning algorithms such as Markov random fields (MRFs) and Boltzmann machines may be used.

[0045] The estimation unit 44 is a processing unit for estimating, based on the submitted data D1, the expansion / contraction state of the base material 9 when the submitted data D1 is printed. The estimation unit 44 performs estimation processing using the estimation model M generated by the learning unit 43. The estimation unit 44 inputs the submitted data D1 acquired by the data acquisition unit 41 to the estimation model M. Then, the estimation model M outputs the expansion / contraction state of the base material 9 corresponding to the submitted data D1. The estimation unit 44 regards the expansion / contraction state output from the estimation model M as an estimation result D5.

[0046] The correction value calculation unit 45 is a processing unit for calculating a correction value D6 based on the estimation result D5. The correction value calculation unit 45 calculates a correction value D6 in a direction for canceling the amount of expansion / contraction of the base material 9 indicated by the estimation result D5. The calculated correction value D6 is input to the operation control unit 46.

[0047] The operation control unit 46 is a processing unit for controlling the operation of the transport mechanism 10 and the four heads 21 to 24. When printing the learning data D2 described above, the operation control unit 46 outputs a command value based on the learning data D2 to the transport mechanism 10 and the four heads 21 to 24. This causes the transport mechanism 10 and the four heads 21 to 24 to operate, and prints the learning data D2 on the printing surface of the substrate 9. On the other hand, when printing the submitted data D1, the operation control unit 46 corrects the command value based on the submitted data D1 by the correction value D6. Then, the operation control unit 46 outputs the corrected command value to the transport mechanism 10 and the four heads 21 to 24. This causes the transport mechanism 10 and the four heads 21 to 24 to operate, and prints the submitted data D1 on the printing surface of the substrate 9.

[0048] <3. Learning process> Next, a description will be given of the learning process executed in the above-mentioned printing device 1. Fig. 5 is a flowchart showing the flow of the learning process. This learning process is executed before printing of the input data D1 to obtain a printed product.

[0049] 5, when performing the learning process, first, the data acquisition unit 41 reads out the learning data D2 from the server 2. Then, the data acquisition unit 41 inputs the learning data D2 to the learning unit 43 and the operation control unit 46 (step S1).

[0050] FIG. 6 is a diagram showing an example of learning data D2. As shown in FIG. 6, the learning data D2 is image data in which a plurality of grid marks 52 are incorporated into an image 51 such as a design or pattern. The plurality of learning data D2 have images 51 that are different from each other. The amount of expansion and contraction of the substrate 9 differs depending on the amount of ink ejected onto the substrate 9. For this reason, it is desirable that the image 51 of the learning data D2 is an image that includes a plurality of regions with different density values ​​or printing rates. This makes it possible to learn the amount of expansion and contraction of the substrate 9 according to the density value or printing rate.

[0051] The grid marks 52 are marks that indicate a predetermined coordinate position in the learning data D2. The grid marks 52 are arranged throughout the learning data D2. The grid marks 52 are arranged at intervals from one another along the transport direction and width direction of the base material 9. In the learning data D2, the grid marks 52 are arranged on the front side of the image 51.

[0052] FIG. 7 is an enlarged view of a portion of learning data D2. As shown in FIG. 7, grid mark 52 in this embodiment is composed of base figure 521 and mark figure 522. Base figure 521 is a rectangular figure filled with white (density value 0%). Mark figure 522 is a black (density value 100%) cross-shaped figure overlapping the front side of base figure 521. When grid mark 52 is arranged on a black background as in the upper region of FIG. 7, the background and mark figure 522 are the same black color, but since mark figure 522 is on white base figure 521, mark figure 522 can be distinguished. Also, when grid mark 52 is arranged on a white background as in the lower region of FIG. 7, since base figure 521 and background are the same white color, the boundary cannot be distinguished, but black mark figure 522 can be distinguished.

[0053] Thus, in this embodiment, the grid mark 52 is composed of a white base figure 521 and a black mark figure 522 overlapping the front side of the base figure 521, so that the mark figure 522 can be identified regardless of the color of the background image 51.

[0054] The operation control unit 46 controls the operations of the transport mechanism 10 and the four heads 21-24 based on the learning data D2. As a result, the learning data D2 is printed on the printing surface of the substrate 9 (step S2). That is, a plurality of grid marks 52 are printed on the printing surface of the substrate 9 together with the image 51. The camera 30 also captures an image of the printing surface of the substrate 9 on which the learning data D2 is printed (step S3). The captured image D3 obtained by capturing the image is transmitted from the camera 30 to the computer 40 and input to the expansion / contraction state measuring unit 42.

[0055] The expansion / contraction state measuring unit 42 measures the expansion / contraction state of the substrate 9 based on the captured image D3 transmitted from the camera 30 (step S4). Specifically, the expansion / contraction state measuring unit 42 extracts the positions of multiple grid marks 52 on the substrate 9 from the captured image D3. Then, based on the positions of these grid marks 52, measures the expansion / contraction state of the substrate 9. As described above, the grid marks 52 are composed of a white base figure 521 and a black mark figure 522. Therefore, the positions of all the grid marks 52 can be recognized regardless of the color or density of the background image 51. Therefore, the expansion / contraction state of the substrate 9 can be measured with high accuracy.

[0056] FIG. 8 is a diagram showing an example of a method for measuring the expansion / contraction state of the base material 9. In the example of FIG. 8, the expansion / contraction state measuring unit 42 measures the interval between adjacent grid marks 52 in the captured image D3. Then, the difference between the measured interval between the grid marks 52 and the interval between the grid marks 52 in the learning data D2 is calculated as the amount of expansion / contraction. That is, in the method of FIG. 8, the change in the interval between adjacent grid marks 52 is measured. The expansion / contraction state measuring unit 42 measures the amount of expansion / contraction in this way for all adjacent grid marks 52 in the captured image D3. As a result, a heat map showing the distribution of the amount of expansion / contraction in the base material 9 is obtained.

[0057] When the base material 9 expands or contracts, the displacement amount of each region on the base material 9 is influenced not only by the amount of expansion or contraction of that region, but also by the amounts of expansion or contraction of other regions, and is a cumulative value of those amounts of expansion or contraction. However, the heat map obtained by the measurement method of Fig. 8 represents the local amount of expansion or contraction of each region of the base material 9. Therefore, the heat map can be easily used as training data in step S5 described later.

[0058] FIG. 9 is a diagram showing another example of a method for measuring the expansion / contraction state of the base material 9. In the example of FIG. 9, the expansion / contraction state measuring unit 42 measures the position of each grid mark 52 in the captured image D3. Specifically, a specific grid mark 52 in the captured image is set as an origin, and the coordinate positions of the other grid marks 52 relative to the origin are measured. Then, the difference between the measured coordinate position and the coordinate position of the grid mark 52 in the learning data D2 is calculated as the expansion / contraction amount. That is, in the method of FIG. 8, the displacement amount of each grid mark 52 in the base material 9 is measured. The expansion / contraction state measuring unit 42 measures such an expansion / contraction amount for all grid marks 52 in the captured image D3. As a result, a vector map showing the distribution of the expansion / contraction amount in the base material 9 is obtained.

[0059] 9 represents the amount of displacement of the coordinate position of each part of the substrate 9. In this case, the estimation result output from the estimation model M, which will be described later, also represents the amount of displacement of the coordinate position of each part of the substrate 9. For this reason, by adopting the measurement method of FIG. 9, it becomes easier to use the estimation result when the correction value calculation unit 45 corrects the coordinate position of the ink ejected onto the substrate 9.

[0060] The expansion / contraction state measuring unit 42 inputs the obtained heat map or vector map to the learning unit 43 as the expansion / contraction state D4 of the base material 9.

[0061] The learning unit 43 performs machine learning by a supervised machine learning program using the learning data D2 input from the data acquisition unit 41 in step S1 as input variables and the stretching state D4 of the base material 9 measured in step S4 as teacher data. At this time, the learning unit 43 prepares an estimation model M for estimating the stretching state of the base material 9 based on the image data. The estimation model M outputs an estimation result of the stretching state based on the input learning data D2. The learning unit 43 adjusts the parameters of the estimation model M so that the estimation result output from the estimation model M approaches the stretching state D4, which is the teacher data (step S5: learning process).

[0062] Thereafter, the learning unit 43 determines whether or not a predetermined termination condition is satisfied (step S6). The termination condition may be, for example, that the difference between the estimation result and the teacher data is smaller than a preset threshold. Alternatively, the termination condition may be that the number of repetitions of steps S1 to S5 reaches a preset threshold. If the termination condition is not satisfied (step S6: No), the computer 40 repeats the above-mentioned processes of steps S1 to S5. In this case, the learning data D2 may be from a different image 51.

[0063] The learning process of steps S1 to S5 is repeated, thereby improving the estimation accuracy of the estimation model M. When the termination condition is eventually satisfied (step S6: Yes), the learning unit 43 terminates the learning process. As a result, a trained estimation model M is generated that can accurately estimate the expansion / contraction state of the substrate 9 when the input image data is printed based on the input image data. The learning unit 43 provides the generated estimation model M to the estimation unit 44.

[0064] Note that a predetermined number of pieces of training data D2 may be treated as one training data set. Then, the training process may be performed for a plurality of training data sets. In this case, while the predetermined number of pieces of training data D2 included in one training data set are being printed, the processes of steps S1 to S5 may be repeated without determining whether the termination condition is met in step S6. Then, when the processes of steps S1 to S5 are completed for all pieces of training data D2 included in one training data set, the determination of whether the termination condition is met in step S6 may be performed. If the termination condition is not met in step S6, the training process of steps S1 to S5 may be performed for another training data set.

[0065] <4. Printing process> Next, a description will be given of the printing process executed in the printing device 1 after the above-mentioned learning process. Fig. 10 is a flowchart showing the flow of the printing process.

[0066] 10, when performing a printing process, first, submitted data D1 to be printed is acquired (step S7: data acquisition step). Specifically, the data acquisition unit 41 reads out the submitted data D1 from the server 2. Then, the data acquisition unit 41 inputs the submitted data D1 to the estimation unit 44 and the operation control unit 46.

[0067] The estimation unit 44 inputs the submitted data D1 to the estimation model M generated by the learning unit 43. The estimation model M then outputs an estimation result D5 of the expansion / contraction state of the base material 9 (step S8: estimation process). The estimation result D5 indicates an estimate of the expansion / contraction state of the base material 9 caused by the ink when the submitted data D1 is printed by the printing device 1. The estimation unit 44 outputs the obtained estimation result D5 to the correction value calculation unit 45.

[0068] The correction value calculation unit 45 calculates a correction value D6 based on the estimation result D5 output from the estimation unit 44 (step S9). This correction value D6 is a control value for finely adjusting the ejection position of the ink droplets with respect to the substrate 9. The correction value calculation unit 45 sets the correction value D6 in a direction that cancels the expansion / contraction state of the substrate 9 indicated by the estimation result D5. For example, when it is estimated that a certain portion of the substrate 9 is displaced to one side in the width direction due to the expansion / contraction of the substrate 9, the correction value calculation unit 45 calculates the correction value D6 so as to correct the ejection position of the ink droplets to the other side in the width direction. Then, the correction value calculation unit 45 inputs the calculated correction value D6 to the operation control unit 46.

[0069] Thereafter, the operation control unit 46 controls the operation of the transport mechanism 10 and the four heads 21-24 based on the submitted data D1 acquired from the data acquisition unit 41 and the correction value D6 acquired from the correction value calculation unit 45. The operation control unit 46 corrects the ink ejection position specified by the submitted data D1 in accordance with the correction value D6. This correction is performed, for example, for each pixel of the submitted data D1. Then, ink droplets are ejected at the corrected ejection position on the printing surface of the substrate 9. As a result, the submitted data D1 is printed on the printing surface of the substrate 9 (Step S10: printing process).

[0070] As described above, in the printing device 1, before printing the input data D1, the expansion and contraction state of the base material 9 due to the ink is estimated based on the input data D1. Therefore, ink droplets can be ejected onto the printing surface of the base material 9 while correcting the ink ejection position in consideration of the estimation result. As a result, a high-quality printout can be obtained with little distortion of the printed image and little misregistration.

[0071] <5. Modifications> Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment.

[0072] <5-1. First modified example> In the above embodiment, the multiple grid marks 52 are arranged at equal intervals in the learning data D2. However, the intervals between the multiple grid marks 52 do not necessarily have to be equal. For example, in a portion of the image 51 where there is a large change in density value or printing rate, the grid marks 52 may be arranged more densely than in other portions.

[0073] <5-2. Second modified example> In the above embodiment, the grid mark 52 is composed of the white base figure 521 and the black mark figure 522. However, the color of the base figure 521 is not necessarily limited to white. Moreover, the color of the mark figure 522 is not necessarily limited to black. For example, the base figure 521 may be black and the mark figure 522 may be white. Moreover, the base figure 521 and the mark figure 522 may be other colors. That is, the grid mark 52 may be formed such that the base figure 521 is formed in a first color and the mark figure 522 is formed in a second color different from the first color. Moreover, the shapes of the base figure 521 and the mark figure 522 may be different from those in the above embodiment.

[0074] <5-3.Third modified example> In the above embodiment, the only information input to the estimation model M in the learning process is the learning data D2. However, in addition to the learning data D2, additional information such as detection values ​​of various sensors in the printing device 1 and the type of substrate 9 may be input to the estimation model M. In that case, in the printing process, the above additional information may be input to the estimation model M in addition to the submitted data D1. In this way, the estimation model M can take the additional information into account and output a more accurate estimation result D5.

[0075] <5-4. Fourth modified example> In the above embodiment, the expansion / contraction state of the substrate 9 is estimated based on the estimation model M generated by machine learning. However, the expansion / contraction state of the substrate 9 may be estimated by other methods. For example, the relationship between the density value or printing rate of each region included in the submitted data D1 and the expansion / contraction state of the substrate 9 may be formulated. Also, the correspondence relationship between the density value or printing rate of each region included in the submitted data D1 and the expansion / contraction state of the substrate 9 may be specified by a table. Then, before printing the submitted data D1, an estimation result indicating the expansion / contraction state of the substrate 9 may be output based on the submitted data D1 and the above formula or table.

[0076] However, when a formula or table is used, the formula or table needs to be modified every time the state of the printing device 1 or the type of substrate 9 changes. The task of modifying this formula or table requires consideration from various perspectives, which places a heavy burden on the user. In contrast, if machine learning is used as in the above embodiment, the estimation model M can be recreated by performing a certain learning process even if the state of the printing device 1 or the type of substrate 9 changes. Therefore, it is easy to respond to changes in conditions.

[0077] <5-5. Other variations> 2, the nozzles 201 are arranged in a row in the width direction in each of the heads 21 to 24. However, the nozzles 201 may be arranged in two or more rows in each of the heads 21 to 24.

[0078] Furthermore, the printing device 1 in the above embodiment has four heads 21 to 24. However, the number of heads provided in the printing device 1 may be one to three, or five or more. For example, the printing device 1 may have a head that ejects spot color ink in addition to the colors K, C, M, and Y.

[0079] Furthermore, the elements appearing in the above-described embodiments and modifications may be combined as appropriate to the extent that no contradiction arises. [Explanation of symbols]

[0080] 1 Printing device 2 Server 9 Base material 10. Conveyor mechanism 20 Printing Department 21 Head 1 22 No. 2 Head 23 3rd Head 24 4th Head 30 Camera 40 Computer 41 Data Acquisition Section 42 Stretching state measurement unit 43 Learning Department 44 Estimation part 45 Correction value calculation section 46 Motion control section 51 images 52 Grid Mark 404 Computer Programs 521 Base Shape 522 Mark Shape D1 Data submission D2 Training data D3 Captured images D4 Stretched state D5 Estimation result D6 Correction value M Estimation Model

Claims

1. A method for estimating an expansion / contraction state of a long strip-shaped substrate in a printing device that ejects ink onto a surface of the substrate while transporting the substrate in a longitudinal direction, comprising: a learning process for generating an estimation model capable of estimating the expansion / contraction state of the substrate by machine learning using learning data, which is image data for learning, as input variables and the expansion / contraction state of the substrate when the learning data is printed by the printing device as training data; A data acquisition step of acquiring input data which is image data to be printed; an estimation step of outputting an estimation result indicating a state of expansion and contraction of a base material due to ink based on the input data before printing the input data; Including, the learning data is image data incorporating a plurality of marks arranged at intervals; In the learning data, the marks are arranged more densely in a portion where the change in density value or the change in printing rate is large than in other portions; In the learning step, a stretching state serving as the teacher data is measured based on positions of the marks on the base material when the learning data is printed by the printing device; In the estimation step, the input data is input to the estimation model, and the expansion / contraction state output from the estimation model is set as the estimation result. The estimation method, wherein the estimation result is an estimation of a two-dimensional distribution of the amount of expansion and contraction of the base material in a transport direction and a width direction.

2. The estimation method according to claim 1, The mark is a base graphic formed in a first color; a mark figure formed in a second color different from the first color at a position overlapping the base figure; An estimation method, comprising:

3. The estimation method according to claim 2, the first color is white; The second color is black.

4. The estimation method according to any one of claims 1 to 3, An estimation method, wherein the training data indicates a change in the distance between adjacent marks.

5. The estimation method according to any one of claims 1 to 3, The teaching data indicates a displacement amount of each of the plurality of marks.

6. The estimation method according to any one of claims 1 to 5, The learning data is image data including a plurality of areas having different density values ​​or printing rates.

7. A printing method using the estimation method according to any one of claims 1 to 6, comprising: After performing the data acquisition step and the estimation step, a printing method comprising: executing a printing step of ejecting ink onto a surface of the substrate while correcting an ejection position of the ink relative to the substrate based on the estimation result.

8. A data acquisition unit that acquires submitted data, which is image data to be printed; A conveying mechanism that conveys a long strip-shaped base material in a longitudinal direction along a predetermined conveying path; a head that ejects ink onto a surface of the base material transported by the transport mechanism based on the input data; a camera that photographs a printing surface of a substrate downstream of the head along the conveying path; an expansion / contraction state measuring unit that measures an expansion / contraction state of the base material based on an image captured by the camera; a learning unit that generates an estimation model capable of estimating the expansion / contraction state of the base material by machine learning using learning data, which is image data for learning, as input variables and the expansion / contraction state when the learning data is printed as teacher data; an estimation unit that outputs an estimation result indicating a state of expansion and contraction of the base material due to the ink based on the input data before printing of the input data; Equipped with the learning data is image data incorporating a plurality of marks arranged at intervals; In the learning data, the marks are arranged more densely in a portion where the change in density value or the change in printing rate is large than in other portions; the expansion / contraction state measuring unit measures an expansion / contraction state to be the teacher data based on positions of the marks on the base material when the learning data is printed; the estimation unit inputs the input data into the estimation model, and regards the expansion / contraction state output from the estimation model as the estimation result; A printing apparatus, wherein the estimation result is an estimation of a two-dimensional distribution of the amount of expansion and contraction in the transport direction and the width direction of the base material.

9. 9. The printing device according to claim 8, a correction value calculation unit that calculates a correction value based on the estimation result; an operation control unit that controls the transport mechanism and the head so as to eject the ink onto the surface of the substrate while correcting an ejection position of the ink with respect to the substrate based on the correction value; The printing device further comprises:

Images